Topic: [FIXED] httpd wont start

[steveo]

Hi all, hoping you can help me.

I cannot start the httpd service:

Code: [Select]

/etc/init.d/httpd start
Starting httpd:               [FAILED]


The httpd error log (tail /var/log/httpd/error_log) says:

Code: [Select]

[Mon Aug 07 23:55:02 2006] [crit] (28)No space left on device: mod_rewrite: could not create rewrite_log_lock
Configuration Failed

There is plenty of space available:

Code: [Select]

df
Filesystem  1K-blocks   Used          Available    Use%     Mounted on
/dev/hda3  76564776   15612052   57063408    22%      /
/dev/hda1  99043        9618          84311          11%      /boot
none          225512      0               225512         0%       /dev/shm


I did install (via yum localinstall) the sme-crontab_manager contrib, but I am not sure this is the cause. I did yum remove and expand the httpd template but httpd still does not start.

I did notice another threads on this:

http://forums.contribs.org/index.php?topic=32490.0
http://forums.contribs.org/index.php?topic=27943.0
[/list]
I have also searched the bug tracker, but can see no resolution.

Any assistance appreciated.

Cheers.

[smeghead]

.. from 10 seconds on Google!

**********************************************************

I did an "everything" install of Fedora Core 2 and it has produced a few small issues. One of them being that Apache gives out when I restart it too much. It turns out to be a memory leak in mod_perl, according to this website. The apache log entries in /var/log/httpd/error_log say:

[crit] (28)No space left on device: mod_rewrite: could not create rewrite_log_lock Configuration Failed

The fix is to run this:

ipcs -s | grep nobody | perl -e 'while () { @a=split(/\s+/); print ipcrm sem $a[1]}'

except where it says "nobody", make sure that matches up to whatever apache runs as. For me it's "apache" not "nobody".

Courtesy of http://inanutshell.us/content/dev.html

In SME apache runs as user www

Remember search and search again!

[cactus]

ipcs -s | grep nobody | perl -e 'while () { @a=split(/\s+/); print ipcrm sem $a[1]}'

except where it says "nobody", make sure that matches up to whatever apache runs as. For me it's "apache" not "nobody".

Changed the user to www and tried the command but it only keeps outputting:

Code: [Select]

usage: ipcrm [ [-q msqid] [-m shmid] [-s semid]
          [-Q msgkey] [-M shmkey] [-S semkey] ... ]
deprecated usage: ipcrm {shm | msg | sem} id ... for minutes on end. Can someone help me fix the syntaxis?

[steveo]

I had the same problem. After a bit of mucking around, I came up with this:

As root run

Code: [Select]

ipcs -s | grep www | cut -d " " -f2 > killthese.txt

To extract the lock file IDs, Then run the following script to knock them off:

Code: [Select]


#!/bin/sh

# open file the for reading
exec 6<killthese.txt
# read until end of file
while read -u 6 dta
do
  echo killing "$dta"
  ipcrm sem $dta
done
#close file killthese.txt
exec 6<&-


This, after some time, managed to kill off all the lock files. Remember to ensure that the httpd service is not running before you run these commands.

Code: [Select]

/etc/init.d/httpd stop

I found I also needed to kill some http processes as well. After I ran these, and killed off all the locks as suggested by the second post (thanks for the info by the way), I still could not start the httpd service. I was getting ssl errors. As a short term solution I had to change the name of the server to get httpd running again. I think I need to delete an auto generated ssl certificate and then rename back to the original name. I'm still trying to figure out how to do this.

Steve

[CharlieBrady]

Hi all, hoping you can help me.

I cannot start the httpd service:

Code: [Select]

/etc/init.d/httpd start
Starting httpd:               [FAILED]


What you are showing here is that you cannot start a *third* httpd service, in addition to the two which are already running, one called "httpd-e-smith" and one called "httpd-admin".

Please read the FAQ and the Developers Guide, both accessible via the Documentation link to the left.

[steveo]

Thanks for the advice charlie. I have changed my server back to its original name - web pages (including server manager) do not display although the service tells me it is running.

Code: [Select]


Stopping httpd-e-smith:                   [  OK  ]
Starting httpd-e-smith:                    [  OK  ]
Stopping httpd-admin:                     [  OK  ]
Starting httpd-admin:                      [  OK  ]


Log confirms this:

Code: [Select]


Aug 10 18:20:32 darkstar httpd-e-smith: Stopping httpd-e-smith succeeded
Aug 10 18:20:40 darkstar httpd-e-smith: Starting httpd-e-smith succeeded
Aug 10 18:21:19 darkstar httpd-admin: Stopping httpd-admin succeeded
Aug 10 18:21:25 darkstar httpd-admin: Starting httpd-admin succeeded


Pages from server are not found, check the httpd/error_log

Code: [Select]


[Thu Aug 10 18:24:52 2006] [crit] (28)No space left on device: mod_rewrite: could not create rewrite_log_lock


Killed the lock files as per previous post. tail /var/log/httpd/error_log

Code: [Select]


[Thu Aug 10 18:36:18 2006] [error] Unable to configure RSA server private key
[Thu Aug 10 18:36:18 2006] [error] SSL Library Error: 185073780 error:0B080074:x509 certificate routines:X509_check_private_key:key values mismatch


There is now a problem with SSL that is preventing the pages from displaying - this would explain how why the pages display with a different server name (not quite sure why the http processes have reported they are running).

I now need to do some research on how ssl is setup (unless anyone here can give me a quick answer).

[steveo]

OK, got the web and admin pages running again.

I deleted the private key and certificate:

Code: [Select]

rm  /home/e-smith/ssl.crt/darkstar.edenmonarolabour.org.crt
rm /home/e-smith/ssl.key/darkstar.edenmonarolabour.org.key


and then ran

Code: [Select]

signal-event domain-modify

If anyone needs to delete those pesky lock files (I needed to to this quite often), here is a script you can use:

Code: [Select]


#!/bin/sh

LOCK_FILE="./killthese.txt"

# First stop the webservers
/etc/init.d/httpd-admin stop
/etc/init.d/httpd-e-smith stop


# Now produce the lock file
ipcs -s | grep www | cut -d " " -f2 > $LOCK_FILE

# open file test.data for reading
exec 6<$LOCK_FILE
# read until end of file
while read -u 6 dta
do
  echo deleting resource "$dta"
  ipcrm sem $dta
done

# close file killthese.txt
exec 6<&-

# delete the data file
rm -f $LOCK_FILE

#now start the webservers again
/etc/init.d/httpd-admin start
/etc/init.d/httpd-e-smith start


[mmccarn]

Same problem.

Deleting the .crt and .key files, signal-event domain-modify, and running steveo's script (thanks!) seem to have solved the problem.

Does anyone know why this happens?

[robin]

Thanks for this script etc this fixed my web problem.

I also had an IMAP problem caused by this same issue
Users were getting "number of connections exceded..." error in thunderbird

you need to kill imap and imaps first then run the commands & script above

Code: [Select]

[root@neptune /]# ps aux | grep imap

root      2206    0.0  0.0  2948  224 ?        Ss   12:04   0:00 runsv imaps
root      2215    0.0  0.0  1508  224 ?        Ss   12:04   0:00 runsv imap

[root@neptune /]# kill 2206
[root@neptune /]# kill 2215


Just thought this may help someone else along the line

Robin

[william_syd]

I also had an IMAP problem caused by this same issue
Users were getting "number of connections exceded..." error in thunderbird

Code: [Select]

[root@c3 ~]# config show imap
imap=service
    ConcurrencyLimit=400
    ConcurrencyLimitPerIP=12
    TCPPort=143
    access=private
    status=enabled
[root@c3 ~]# config show imaps
imaps=service
    ConcurrencyLimit=400
    ConcurrencyLimitPerIP=100
    TCPPort=993
    access=public
    status=enabled
[root@c3 ~]#


[robin]

Thanks William,

I did the "ConcurrencyLimit" stuff first but the imap still wasn't working because of the security cert issue.
A friend pointed me to this thread and it worked but perhaps the script could be modified to stop imap and imaps?

would it be like this?

Code: [Select]


#!/bin/sh

LOCK_FILE="./killthese.txt"


# First stop the webservers
/etc/init.d/httpd-admin stop
/etc/init.d/httpd-e-smith stop

#Now stop the imap servers
/etc/init.d/imap stop
/etc/init.d/imaps stop

# Now produce the lock file
ipcs -s | grep www | cut -d " " -f2 > $LOCK_FILE

# open file test.data for reading
exec 6<$LOCK_FILE
# read until end of file
while read -u 6 dta
do
  echo deleting resource "$dta"
  ipcrm sem $dta
done

# close file killthese.txt
exec 6<&-

# delete the data file
rm -f $LOCK_FILE

#now start the webservers again
/etc/init.d/httpd-admin start
/etc/init.d/httpd-e-smith start

#Now start the imap servers again
/etc/init.d/imap start
/etc/init.d/imaps start

I added the imap start and stop, but do we need something in this line ?

Code: [Select]

# Now produce the lock file
ipcs -s | grep www | cut -d " " -f2 > $LOCK_FILE

I notice the grep is just www?
does it need to look for imap in there also?

New to this and would like to use this as a learning experience, Thanks

[cfmunster]

I banged my head on this problem for awhile today. I whacked my httpd install while trying to set up SugarCRM per this thread:

http://forums.contribs.org/index.php?topic=35112.0

I finally did all the things recommended here, but I also removed eaccelerator.so and its settings from my php.ini file, which finally did the trick and got me running again. I might leave it off foe a bit and try and install it again later, but I don't have high usage, so it isn't a big deal.

[jdavey]

Same problem...

This was on our PDC dual Xeon s with 4 GB of RAM. 40 users. Ridiculous uptime on this box.

As stated by others: Deleting the .crt and .key files, signal-event domain-modify, and running steveo's script have solved the problem.

I'll go check the Bug tracker to see if there are issues reported there.

[robwellesley]

I had no 'luck' until I stopped doing

/etc/init.d/httpd start

and did

svc - u /service/httpd-e-smith

[cactus]

I had no 'luck' until I stopped doing

/etc/init.d/httpd start

and did

svc - u /service/httpd-e-smith

That is because the service control has changed, the /etc/init.d is deprecated, you should use

Code: [Select]

sv t httpd-e-smith (restart) instead.